| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
uta1,2
ák1
í Popelá11 Institute of Experimental Medicine, Academy of Sciences of the Czech Republic, 142 20 Prague; 2 3rd Medical Faculty, Charles University, 100 00 Prague, Czech Republic
Submitted 30 December 2002; accepted in final form 21 August 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
25% stronger than the response to any other sound (tone, noise, and other calls); these neurons were selective for chirp or whistle, and no unit preferred chutter or purr. Neuronal activity provided information about the spectrotemporal patterns of the calls. Peristimulus time histograms (PSTHs) reflected the energy of the near-characteristic frequency band, and the population PSTH reliably matched the sound envelope for calls characterized by one or more short impulses (chirp, purr, and chutter) but did not exactly fit the envelope for whistle—a slow-modulated and relatively long call. Calculations based on firing rates indicated the approximate positions of the main spectral peaks but did not always reflect their relative magnitude. The time-reversed version of whistle elicited on average a weaker response than did the natural whistle (by 24%), but there were neurons with a significantly stronger response to the natural ("forward-selective," 30%) as well as to the time-reversed whistle ("reverse-selective," 15%). This study does not prove the existence of units selectively responding to animal calls, but it provides evidence for the encoding of the spectrotemporal acoustic patterns of vocalizations by IC units. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
A traditional concept in the perception of such signals is based on the existence of highly specific neurons, so-called call detectors. The first studies performed in the auditory cortex of awake squirrel monkeys suggested that neurons might exist in the cortex that extract specific features of calls, similarly to the visual cortex performing a feature extraction (Manley and Müller-Preuss 1978
; Newman 1978; Newman and Symmes 1974; Newman and Wollberg 1973; Winter and Funkenstein 1973; Wollberg and Newman 1972). The existence of specific cells—"call detectors"—was not confirmed in these studies, and the authors later came to the conclusion that the pattern discrimination of a complex sound can be accomplished by a functional ensemble of neurons (Pelleg-Toiba and Wollberg 1991). Later, Rauschecker et al. (1995) found a preference for increasingly complex stimuli in neurons in the superior temporal gyrus of anesthetized rhesus monkeys and suggested that the lateral belt areas of the monkey auditory cortex may form an important stage for the processing of communication sounds. Similarly, Wang et al. (1995) described a representation of behaviorally important and spectrotemporally complex species-specific vocalizations in the primary auditory cortex of anesthetized common marmosets. In their view, the representation of vocalizations is carried out by dispersed and synchronized cortical cell assemblies that correspond to each individual vocalization in a specific and abstracted way.
In contrast to the large number of studies that have investigated the role of individual cortical areas in the processing of animal vocalizations, the subcortical nuclei have attracted less attention in this regard. Creutzfeldt et al. (1980) studied the thalamocortical transformation of responses to complex auditory stimuli in nonanesthetized guinea pigs. They concluded from their results that the responses of medial geniculate body (MGB) cells represent more components of a call than do cortical cells, even if the 2 cells are synaptically connected. The responses of MGB neurons to species-specific vocalizations in guinea pigs were also the subject of a study by Tanaka and Taniguchi (1991). These authors observed a low responsiveness of MGB neurons to vocalizations; however, the responses to vocalizations displayed discharge patterns that were not possible to predict from the properties of their responses to pure tones. Only a few studies have been designed to investigate the responses of neurons in the inferior colliculus (IC) to vocal stimuli (Aitkin et al. 1994; Poon and Chiu 1997). The aim of the study by Aitkin et al. (1994), performed in anesthetized cats, was to gain information about the differential coding properties of neurons in 3 subdivisions of the IC: the central nucleus (CNIC), external cortex (ECIC) and dorsal cortex (DCIC). Feline vocal stimuli were found to be more effective in terms of higher firing rates than white noise or pure tone stimuli at the characteristic frequency (CF) in the ECIC and DCIC compared with that in the CNIC. There were no units that responded exclusively to one vocal stimulus, but a high proportion of units in the ECIC responded strongly to broadband stimuli, and some of these showed a clear preference for one vocal stimulus over another.
In the present study we focused on guinea pig communication sounds. The guinea pig has a large repertoire of vocal communication calls (11 distinct calls according to Harper 1976). These calls fundamentally differ in their spectrotemporal features (Syka et al. 1997), but each call is stereotyped from one to another vocalization and also from one animal to another. We analyzed the responses of IC neurons to 4 main types of guinea pig calls. Our goals were to determine how species-specific vocalizations are represented in the IC and what information about the spectral and temporal characteristics of these complex signals is provided by the IC units to higher structures of the auditory pathway.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Experiments were performed on 24 adult, healthy, pigmented female guinea pigs weighing 300–500 g. The care and use of animals reported in this study were approved by the Ethics Committee of the Institute of Experimental Medicine and followed the guidelines of the Declaration of Helsinki. Animals were anesthetized with an intramuscular injection of 1 ml/kg of a mixture of ketamine (Narkamon 5%, Spofa) and xylazine (Rompun 2%, Bayer) at a ratio of 2:1, which corresponds to a dose of 33 mg/kg of ketamine and 6.6 mg/kg of xylazine. Supplementary injections of one half of the original dose of the ketamine–xylazine mixture were administered every hour to maintain a sufficient level of anesthesia.
The skin and underlying muscles on the skull were retracted to expose the dorsal cranium between points bregma and lambda. A small hole (diameter about 5 mm) was made by a trephine in one side of the skull above the IC and the dura mater was removed. The animal's head was rigidly held in a stereotaxic apparatus by a U-shaped holder, which is fixed at its base to the skull by 2 screws and secured by acrylic resin. This type of fixation enabled the animal's head to be free for electrode penetration and for free-field acoustical stimulation. A DC-powered electric heating pad maintained a rectal temperature of 37–38°C.
Acoustic stimulation
Animals were placed in a soundproof anechoic room and acoustical stimuli were delivered in free-field conditions from a 2-way loud-speaker system (Tesla ARN 5614 and Motorola KSN-1005) placed 70 cm in front of the animal's head. The acoustic system was calibrated with a B&K 4133 microphone, placed in the position of the animal's head and facing the speakers. The frequency–response curve was relatively flat and varied by less than ±9 dB between 0.15 and 45 kHz.
Two types of simple stimuli were applied: pure-tone pips or broad-band noise (BBN) bursts (both of 100-ms duration with 3-ms rise/fall times presented with the repetition rate of 1 Hz). Sound stimuli were generated by a sound-wave generator (Hewlett–Packard HP33120A) and shaped with a custom-made electronic gate. This equipment was used to reduce the intensity of the stimuli from maximal output by minimal steps of 1 dB.
Four typical vocalization calls were chosen from the large variety of guinea pig natural calls (11 distinct calls according to Harper 1976). Calls were previously tape recorded from spontaneously vocalized female guinea pigs (age 2–24 mo) in a sound-attenuated room. During the experiment calls were presented from a DAT recorder (Sony DTC-55ES) by the electronic gate. The temporal and spectral parameters of the calls and their variability were previously described in Syka et al. (1997). One representative of each call, shown in Fig. 1, was selected for this study.
|
Purr (Fig. 1A) consists of a series of regular low-frequency impulses (fundamental frequency around 300 Hz). The most complex sound is whistle (Fig. 1B), which is a long-lasting frequency- and amplitude-modulated sound consisting of many harmonics in a wide frequency range. Chutter (Fig. 1C) is a sequence of irregular noise bursts and chirp (Fig. 1D) is an isolated brief acoustic impulse with a harmonic structure. The stimuli were presented once every 2.9 s for chutter and purr, 2.5 s for chirp, and 2.2 s for whistle. Unless specified, all vocalization responses reported in this study were obtained at a maximal effective sound level of 75 dB SPL (sound pressure level dB re 20 µPa). The time-reversed whistle was generated by reversing the time course of the natural whistle call. All other parameters of the call, such as the sound level or the repetition rate, were preserved.
Recording of neuronal activity
Neuronal responses were recorded during a 30- or 60-s period when either simple sounds or vocalization signals were presented. Extracellular unit activity was recorded with a glass micropipette filled with 3 M KCl. The electrode was inserted into the IC in a dorsoventral direction through the cortex, using an electronically controlled microdrive with 1-µm steps. One single track or several parallel penetrations separated by 300–500 µm were made in one or 2 frontal or sagittal planes. As a search stimulus, BBN bursts or tone sweeps were used.
The signal from the electrode was amplified by a WPI DAM 60 differential amplifier and band-pass filtered in the range of 300 Hz to 10 kHz. Then the signal was transmitted by a CED 1401plus interface into a PC computer running the Spike2 program, where the activity was saved and later analyzed. Typically, the activity of only a single unit was recorded from one microelectrode, but in some experiments a unit pair was recorded from one electrode at the same time. In this case, the recorded neuronal activity was processed with Spike2 software to discriminate the individual units in the record according to their spike shapes. Only reliably discriminated units were included in the data set.
Data analysis
The following parameters of the response were evaluated in the process of data analysis:
) and Rees and Palmer (1989) for 2 sounds consisting of several phrases: chutter and purr. The values of the VS were evaluated from the entire response to chutter (i.e., from the response to all 5 phrases of this sound) and from the response to the last 11 phrases of purr. The significance was evaluated using the Rayleigh statistics (Barbour and Wang 2002; Gaese and Ostwald 1995). ; Wang et al. 1995) by aligning individual PSTHs (bin width 5 ms) according to the CF (bin width 0.05 kHz) of individual unit (y-axis). If 2 or more units had the same CF, the values were dispersed into adjacent bins, but never changed by more than 0.1 kHz. Finally, the 3D plots were discriminated at the level of 80 spikes/s; that is, only a driven rate over this value is represented by a black dot in the presented figures; all other bins are white. ) of all units having a CF within a 0.35-octave-wide frequency window moved in 1/8-octave steps. Only points calculated from 3 or more units were included into the profile. The response was calculated from a time window shifted by 10 ms relative to the call to compensate for the response latency. All statistical tests were performed in Prism software.
Histological control
At the end of the experiments, the electrode was left in the deepest position of the last penetration within the IC and carefully broken just above the surface of the brain. The broken tip of the electrode clearly marked the electrode track when resting in the perfused brain for several days. The guinea pigs were killed with an overdose of 1–2 ml of pentobarbital (Pentobarbital, Spofa, 50 mg/ml) and perfused with 10% formaldehyde. The brains were sectioned on a freezing Reichert microtome (slice thickness 40 µm) and stained with cresyl violet. In animals in which only one electrode penetration was made or in which multiple tracks were made in one or more frontal planes, the brain was cut in the frontal plane. Similarly, sections were cut in the sagittal plane in guinea pigs with several electrode penetrations in the sagittal planes. Individual electrode tracks within the IC were subsequently reconstructed from histological sections, and the position of individual neurons was assessed according to the coordinates indicated on the electronically driven microdrive.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
Unit responsiveness to simple sounds
The majority of recorded neurons (92%) were spontaneously active with an average spontaneous firing rate of 12.3 ± 16.8 spikes/s. All but 4 units responded to pure-tone stimulation. CF was evaluated for all units responding to pure tones and ranged between 60 Hz and 25 kHz. The distribution of neurons according to CF is presented in Fig. 2B. All neurons not driven by pure tones responded to stimulation with BBN and/or with vocalization sounds. Based on the shape of the PSTH obtained for stimulation by a pure tone at the CF, 2 types of neuronal responses were distinguished: one-third of neurons (33%) responded only at the stimulus onset (onset units), whereas two-thirds of neurons (67%) responded during the whole stimulus presentation (sustained units).
Stimulation with BBN evoked an excitatory response in 81% of neurons. Five percent of neurons displayed an inhibition of their spontaneous activity during BBN presentation, and the remaining 14% of units did not respond to BBN at all (mostly low-CF neurons with a CF below 1 kHz). Neurons with excitatory responses to BBN were mostly sustained units (91%); only 9% of IC neurons responded to BBN stimulation with an onset response. The frequency tuning of individual IC neurons was characterized by the Q10 value, which ranged from 0.1 to 16. In the majority of neurons (75%), Q10 values were between 1 and 5. As a rule, the Q10 value increased with increasing CF. These characteristics of the responses to simple stimuli are fully compatible with those reported in a previous extensive study of the response properties in the IC of the guinea pig (Syka et al. 2000).
Unit responsiveness to vocalization calls
A total of 124 IC neurons were held long enough to record their responses to all 4 calls. The distribution of IC neurons according to their responsiveness to vocalizations is demonstrated in Fig. 3A. More than half of IC neurons (55%) responded to all 4 types of vocalization signals by either an excitatory or an inhibitory response; 23% responded to 3 vocalizations; 16% of neurons reacted to 2 vocalizations; and only a small portion of neurons responded to only one call (3%) or did not respond to any call (3%). All neurons that were driven by only one call or were not driven by vocalizations at all responded to stimulation with pure tone and/or with BBN. The responsiveness of IC neurons expressed as the percentage of units responding to individual calls (Fig. 3B) ranged from the lowest responsiveness to purr (74% of neurons) to the most efficient, chutter (87% of neurons).
|
When the strength of the response, evaluated as the firing rate over the duration of the stimulus, was compared between the calls and simple sounds (tones, BBN), 37% of units gave a stronger response to one or more calls than was the maximal response to tone or BBN. The majority of these units that preferred vocalization signals gave the strongest response to chirp (67%); whistle was the most efficient in 23% of neurons, chutter in 6%, and purr in 4%. In 15% of all units, the response to the most efficient call was 25% stronger than the response to the second most efficient sound (either tone, BBN, or another call); 80% of these units were selective for chirp, 20% for whistle, and no units preferred chutter and purr. The stronger response to chirp and whistle corresponds to the results shown in Fig. 2A, which demonstrate that the average driven response was stronger in the case of chirp and whistle than in chutter and purr.
The ability of a neuron to respond to an individual call was evaluated in relation to its response pattern to a pure tone, classified as either an onset or sustained response pattern (Fig. 3C). The figure demonstrates that the efficacy of stimulation with vocalization calls did not depend on the pure-tone response pattern; that is, the portion of neurons responding to individual calls was similar for neurons with an onset or sustained PSTH response pattern.
The responsiveness to vocalization signals was also compared with the CF of individual neurons. The results of this analysis, presented in Fig. 4, showed that only a relatively small portion of low-CF neurons (with CF <1 kHz) responded to stimuli with a high-frequency content, such as whistle (41% is a significantly lower portion than in any other CF group, P < 0.0001, 
2 test). On the other hand, high-CF neurons (CF >4 kHz) responded significantly less frequently to purr (P < 0.005, 2 test), i.e., to a call with a predominantly low-frequency content. The agreement between a lower percentage of responding neurons in a particular CF-range and the lack of call energy in that frequency range suggests that the responsiveness of IC neurons to vocalizations is dependent on the relationship between the CF of a neuron and the frequency spectrum of the call.
|
Spectrotemporal discharge pattern of the population response
To evaluate the role of the spectrotemporal features of vocalization calls in the response of IC units, population response maps were constructed from the PSTHs of individual units and compared with the spectrograms of vocalization sounds (Fig. 5). A black dot in the population response map represents an elevated discharge of a unit with a given CF (ordinate) occurring during or after vocalization presentation (abscissa). Population response maps for all 4 tested calls are displayed in Fig. 5. All population response maps demonstrate the relationship between the spectrotemporal patterns of the stimuli and the neuronal responses: the areas of elevated neuronal activity match the areas of energy in the spectrograms. The main difference between the stimuli and the population responses is the wider CF range of discharging units than the corresponding frequency range in the stimulus (i.e., a unit with a CF equal to a frequency absent in the stimulus can also respond to a call if the stimulus spectrum extends into the unit's receptive field).
|
This factor is more prominent in the responses of high-CF neurons to low-frequency sounds such as purr or chutter (Fig. 5, A and C) than in the responses of low-CF neurons to sounds of higher frequencies (e.g., the weak response of neurons with CF <1 kHz to whistle; see Figs. 4 and 5B). This pattern is apparently associated with the shape of the tuning curves at higher intensities, where the low-frequency tail enables responses of high-CF neurons to low-frequency sound content, but the sharp slope of the high-frequency border of the tuning curve does not allow the response of low-CF units to sounds of higher frequencies. The spectral and temporal aspects of the encoding are analyzed in detail in the following sections.
Representation of temporal features
Purr and chutter are characterized by a rhythmic repetition of several phrases. The frequency of the regular repetition of phrases (i.e., the vocalization frequency) (Wang et al. 1995) is expressed in the neuronal response pattern. The spectrum of the population PSTH (Fig. 6) indicates the vocalization frequency of purr as well as of chutter (Fig. 8). The synchronization of the neuronal firing to the vocalization frequency is also detectable from PSTHs of individual units (Fig. 7), where 64% of units show significant synchronization to purr (Rayleigh statistics of the VS, P < 0.001) and 61% of units to chutter. The synchronization to the vocalization stimuli is more frequent in low-CF units: 86% of units with a CF below 4 kHz are significantly synchronized to chutter, but only 48% of units with a CF above 4 kHz are synchronized. In the case of purr, the statistics show that 75% of units with a CF below 4 kHz are significantly synchronized to the stimulus, whereas only 54% of units with a CF above 4 kHz follow the stimulus.
|
|
|
A comparison of population PSTHs (i.e., average PSTHs) and sound envelopes (Fig. 6) shows a good agreement in their shapes for sounds characterized by one or more short phrases, i.e., purr, chutter, and chirp (Fig. 6, A, C, and D). Neuronal responses to these calls follow the energy of the sound, and the magnitudes of the peaks in the population PSTHs were approximately proportional to the peak magnitudes in the sound envelope. The similarity between the population PSTH and the sound envelope can be quantified by the correlation coefficient, which was 0.93 for chirp, 0.81 for chutter, and 0.71 for purr. These values, calculated for the average responses of the entire population of neurons, were below the maximal single neuron values for chirp and purr. The response pattern of the neuron showing the maximal correlation to the sound envelope of chirp had a correlation coefficient of 0.97. The maximum correlation coefficient for purr was 0.80. In the case of chutter, the highest single unit correlation coefficient (r = 0.71) was lower that the correlation coefficient of the entire population. When the units having the highest correlation coefficients to the sound envelope were grouped together, the correlation coefficient between the average response of the group and the sound envelope reached an even higher value than that of the best unit. The maximal value of the correlation coefficient was reached by a group of 5 neurons with the highest correlation coefficients in the case of chirp (r = 0.98), 10 neurons in the case of purr (r = 0.90), and 40 neurons in the case of chutter (r = 0.88). Similarity of values calculated for the best unit, the entire population, and the optimal subpopulation suggests that the information about the peaks in the envelope of the calls could be available at the level of single units or a small neuronal pool.
Different results were obtained for whistle. The correlation coefficient between the population PSTH and the sound envelope was 0.77, but there were some striking differences: the shape of the population PSTH fundamentally differed from the monotonically increasing envelope of the call (Fig. 6B), as the population PSTH was multimodal with 2 main peaks: one at the beginning of the sound (about 150 ms) and the second during the sound (about 320 ms).
Subpopulation PSTHs (Fig. 9), calculated only for units from a limited CF range, demonstrate that there are different sources of these peaks. The early peak represents the onset activity of units with CFs lower than 7 kHz (Fig. 9A). These units have no enhanced activity at about 320 ms, and the sustained pattern is adequate for the sustained character of the frequency range <7 kHz in the sound. The monotonically increasing pattern of the sound energy does not exactly correspond to the response pattern (r = 0.48) because the neuronal activity decreases from the elevated onset part to a constant level in the second half of the response.
|
The 2nd peak in the population PSTH reflects the onset activity of units with CFs above 8 kHz (Fig. 9B). These high-CF units weakly reacted or remained silent until about 320 ms because there is almost no energy in the appropriate frequency component in whistle during this time interval. They began responding at about 320 ms because of the extension of the stimulus spectrum above 8 kHz. There is one more peak in the last part of the response that corresponds with the higher energy content before the end of the stimulus. The correlation coefficient between the relevant frequency component and the response pattern was 0.80.
The differences in subpopulation PSTHs indicate the importance of energy in the near-CF frequency band for the neuronal response. The relevance of the near-CF frequency band is further demonstrated in individual neurons in Fig. 10, where various patterns of the neuronal responses (PSTHs) are compared with the sound obtained by the band-pass filtering of whistle for 4 units with different CFs. The correlation coefficient between the response and the near-CF component is higher than the correlation coefficient between the response and the entire vocalization (e.g., 0.66 vs. 0.45 for the unit in Fig. 10C). These data clearly demonstrate the importance of the near-CF frequency for generating the unit response and suggest that the unit response to whistle stimulation is based on the relationship between the stimulus spectrum and the neuronal excitatory and inhibitory response fields.
|
The variability of the response patterns is further emphasized by the fact that units with close or identical CFs can respond in different ways, as shown in Fig. 11. The response of the unit shown in Fig. 11A (top) is positively correlated (r = 0.83) with the envelope of the near-CF component, but the unit shown in Fig. 11A (middle) displays a negative correlation (r = —0.44). Similarly, the response patterns of units shown in Fig. 11B have nonzero correlation coefficients to the near-CF component: the top PSTH is positively correlated (r = 0.61) and the bottom PSTH is negatively correlated (r = —0.06). For both neurons displaying a negative correlation, the shape of the response corresponds with the nonmonotonic rate-level function observed when stimulated by a tone at the CF.
|
Responses to whistle and reversed whistle
In some experiments, whistle was presented in a temporally reversed form with the aim of evaluating the selectivity of neuronal responses for a given temporal pattern. This procedure changes the temporal features of the sound, but preserves the spectral characteristics.
In Fig. 12, a comparison of the responsiveness measured by the driven rate to the natural and reversed calls is shown. Two of 47 units reacted in a different way to the stimuli, i.e., by an excitatory response to one stimulus and an inhibitory response to the other, although both these neurons gave only weak driven responses and the difference between them was not significant (Mann–Whitney test, P > 0.05).
|
The driven firing rates obtained for whistle and reversed whistle in the same unit were positively correlated (R2 = 0.69). The slope of the regression line (0.76) was significantly different from one (P < 0.003), and the slope <1 indicates a weaker average response to the temporally reversed stimulus. The responses of 21 units (45% of evaluated neurons) differed significantly (Mann–Whitney test, P < 0.01): 14 units gave a significantly stronger response to whistle (the slope of the regression line was 0.59 for these units), and 7 reacted with a significantly stronger response to the reversed whistle (the slope of the regression line was 1.41 for these units).
Representation of spectral features
A comparison of the spectrotemporal acoustic and response patterns demonstrated a strong dependency of the firing pattern of the units on the spectral composition of the call. To understand what information about the spectral features is coded by neuronal firing, the neuronal representation of the spectral characteristics was analyzed using a rate versus CF profile, which was compared with the short-term sound spectrum.
In Fig. 13, rate versus CF profiles are shown for the whistle at 3 different times (A–C) and for purr (D), chirp (E), and chutter (F). Figure 13A shows the situation at the beginning of the whistle (150–260 ms). The correlation coefficient between the short-term sound spectrum and the appropriate rate versus CF profile is 0.38. The peaks in the rate versus CF profile indicate the positions of the 2 main spectral peaks. The magnitudes of the rate peaks do not exactly fit the spectrum of the sound because the rates of units with a CF corresponding to the fundamental frequency are slightly lower than that of the 1st harmonic; this is apparently caused by inhibition, that is, by negative driven rates of a part of the low-CF units (Fig. 10A). The range of the 2nd and 3rd harmonics is represented by one peak, and there is a near-zero level of driven activity in the high-CF region (>7 kHz), which corresponds to the absence of any high-frequency component in the sound at this time.
|
Figure 13B shows a different situation, which is constructed for the middle part of whistle (260–370 ms). Although spectral peaks are reflected in the rate-CF profile, the magnitudes of the rate peaks do not correspond to the magnitudes of the spectral peaks. Even though the dominant frequencies are the fundamental and 1st and 2nd harmonics, the strongest response is at the frequency of the 3rd harmonic (about 8 kHz), and the high-CF area is also elevated more than would be expected from the short-term spectrum of the sound; the correlation coefficient is only 0.09. This discrepancy could be explained by the onset reaction of high-CF neurons, which exceeds the sustained activity of lower-CF neurons.
Figure 13C is constructed for the latest phase of whistle (430–540 ms) when the maximal peak in the rate-CF profile indicates the dominant spectral peak of the fundamental frequency in the short-term sound spectrum; the correlation coefficient is 0.35.
The rate-CF profile for chirp (Fig. 13E) also indicates the positions of the main spectral peaks, with isolated peaks of the fundamental and 1st harmonics frequency and one peak for the 2nd and 3rd harmonics; the correlation coefficient is 0.42. A somewhat different situation is seen for the other 2 sounds (Fig. 13, D and F), where some local spectral peaks are intensified in the rate-CF profile and create dominant elements. In both cases, a low level of high-CF driving rates reflects the weak high-frequency component of the calls with correlation coefficients of 0.54 for purr and 0.69 for chutter.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
). Because it is known that in the auditory cortex of many mammals, particularly primates and bats, specialized areas exist devoted to the processing of vocalizations (Ohlemiller et al. 1996; Rauschecker et al. 1995), it remains to be elucidated to what extent the specialized processing of species-specific vocalizations has a subcortical participation.
The representation of guinea pig communication sounds in the neuronal population within the IC was examined in the present study. The CF range of the recorded neurons covered the entire spectral range of the communication sounds that were employed. The set of units used in this study was comparable to the extensive mapping of IC neurons done by Syka et al. (2000), under the same anesthetics, in terms of the basic parameters of unit responses such as for example latency, threshold, spontaneous activity, and Q10.
The major findings can be summarized in 6 points: 1) each of the guinea pig vocalizations evoked responses in about 80% of units; 2) the selectivity of IC units for individual calls was very low as the majority of units responded to all calls; 3) the spectrotemporal discharge pattern of the IC neuronal response was qualitatively similar to the spectrotemporal pattern of the species-specific vocalization; 4) the temporal envelope of the call was coded by the firing rate, except for the slow modulation of whistle; 5) the main spectral peaks of the call spectrum were expressed in the firing rate of IC units; 6) the average response of IC neurons was significantly stronger to the natural than to the reversed whistle, but there were also neurons that significantly preferred the time-reversed whistle.
Coding of spectrotemporal features
The results of the study demonstrate that the presence of highly specialized neurons in the IC with the ability to detect specific features of a call is unlikely. Although the existence of such highly specific units cannot be conclusively ruled out, it is more likely that a population of IC neurons represents the spectrotemporal features of a call.
The population of neurons in the IC codes the acoustical pattern of a call in such a way that the presence or absence of neural responses is a consequence of the tuning properties of the IC units and of the spectrotemporal acoustical pattern of the sound.
The PSTHs of individual IC units demonstrated a close relationship to the energy of the near-CF frequency component of the call. This observation suggests a dominant role for the near-CF frequency band. Individual PSTHs (and especially average PSTHs) reliably copy the envelope of calls characterized by one or more short phrases (i.e., chirp, chutter, and purr). In calls containing more than one component (phrase), such as purr and chutter, the acoustical patterns of these components are very stereotypic. Also, the individual peaks in the response pattern are very stereotypic with the exception of the 1st (onset) peak in the response to purr, which is enhanced more than would be expected from the acoustical pattern. Some variability in the peak amplitude of the sound is reflected in the response, given that a stronger response is equivalent to a greater intensity of the phrase.
The response seems to omit slow modulation of the sound envelope. This phenomenon is present mainly in the response to whistle, in which the sustained character of the response indicates just the presence of energy, but the slow changes in the sound envelope are not reflected in the modulations of the firing rate. This inability of units to follow the slow fluctuations in the envelope corresponds well with the weak synchronization between neuronal discharge and sound envelope as seen for sinusoidal amplitude-modulated tones at low modulation frequencies. The modulation transfer function of IC units typically has a band-pass character as shown by Rees and Moller (1983) in rat and later by Rees and Palmer (1989) in guinea pig. Our findings also correlate with results obtained in the auditory cortex of the marmoset (Nagarajan et al. 2002), where degradations in the temporal envelope performed by low-pass filtering of the temporal envelope at 4 and 10 Hz dramatically diminished the synchronized response.
The rate versus CF plots identified some of the major spectral components of the sound. The spectral profile calculated from the response did not always exactly fit the short-term spectrum, but it clearly marked the position of the main spectral peaks. There were 2 main discrepancies: in the relative magnitudes of the spectral peaks and in the representation of the higher harmonics. In some cases, at first the magnitudes of the sound spectral peaks were not exactly matched by the magnitudes of the peaks in the response profile. The magnitudes of the peaks in the low-CF region are lowered by inhibition, which predominantly occurred in low-CF units, and also the pattern of temporal modulation may affect the spectral profile because an onset firing of units within a particular CF range can be greater than a sustained firing within another CF range, irrespective of the magnitudes present in the sound spectrum (Fig. 13B).
Later, higher harmonics are not represented by several individual peaks but rather by a complex peak. This could be caused by fairly broad tuning curves at the sound intensity used for vocalization to produce the responses and also by the calculation method employed, the averaging procedure of which used a 0.35-octave-wide window and thus limited the spectral resolution.
Although there are limits to the rate representation of fine detail in the sound spectrum, Nagaranjan et al. (2002) showed in the auditory cortex of anesthetized marmosets that response magnitudes are relatively insensitive to the fidelity of the spectral envelope characteristics of a call, and Shannon et al. (1995) demonstrated that speech intelligibility is preserved as long as temporal envelope cues are present, despite spectral degradation to as few as 4 broad frequency bands.
Whistle versus reversed whistle
The importance of the temporal structure of the call is demonstrated by the fact that whistle evoked a stronger response than did the artificial, time-reversed whistle. Not every unit responded in this manner; some units reacted more strongly to the reversed sound, but the average response calculated over the population was significantly weaker for the reversed sound. The preference of neuronal responses for natural communication sounds over artificial, time-reversed sounds has been reported in cat (Gehr et al. 2000), primates (Wang and Kadia 2001; Wang et al. 1995), songbirds (Doupe and Konishi 1991; Margoliash 1983), and bat (Esser et al. 1997).
Our data demonstrate that a preference for natural calls over reversed calls is already present in the guinea pig at the subcortical level, but this preference is weaker than in the case of the auditory cortex in the marmoset (Wang et al. 1995). These authors characterized about 76% of neurons in the primary auditory (AI) cortex of the anesthetized marmoset as "forward selective," that is, responding more strongly (measured by the synchronized discharge rate) to a natural call than to its time-reversed form. Weaker selectivity was reported by Gehr et al. (2000), who analyzed the responses of AI neurons in the anesthetized cat to meow and time-reversed meow. They divided the response into 2 parts: an onset part, occurring within 50 ms after beginning of the call; and a sustained part, represented by the activity occurring later. They reported a larger onset response to the natural meow than to the time-reversed meow (in a ratio of about 2:1), but no significant differences between the natural and artificial calls in the sustained part.
The importance of subcortical processing of behaviorally relevant sounds has been reported in the mustached bat, where it is assumed that pulse + echo combination-sensitive responses originate in the MGB (Wenstrup and Grose 1995) or even in the inferior colliculus or below (Wenstrup et al. 1999). Also significant are the results of Aitkin et al. (1994) and Tanaka and Taniguchi (1991) showing that responses to vocalizations are not always predictable from the responses to pure-tone stimuli.
The possibility that the weaker response to temporally manipulated sounds is attributable to their lack of behavioral relevancy is supported by the study of Wang and Kadia (2001) showing that cortical neurons in cat responded similarly to marmoset natural and time-reversed communication sounds but marmoset cortical units preferred the natural call over the time-reversed call. The origin of the stronger response to whistle may be determined by the call features. There are 2 main features of the whistle: an amplitude modulation (AM) of the sound and also a frequency modulation (FM) determined by the increasing frequencies of all harmonics. There is a possibility that the preference for natural whistle could be based on the preference for rising amplitude versus falling amplitude, for rising frequency versus falling frequency, or for a combination of these. This idea is supported by studies of neuronal responses to amplitude or frequency-modulated sounds. Chiu and Poon (2000) reported that more than half of the cells in the IC of urethane-anesthetized rats could be considered as "AM-sensitive" with a preference for the rising phase of the AM. Neuert et al. (2001) reported asymmetry in the responses to damped and ramped sinusoids in the IC of anesthetized guinea pigs, where all units displayed significant asymmetry in the discharge rate for at least one time constant of the AM. The relevance of FM was demonstrated by the study of Kao et al. (1997), who made a reliable prediction of the response to a rat vocalization call based on the response to an FM tone.
Diversity of calls
The repertoire of guinea pig communication sounds is a discrete set of calls, but it is a very diverse group with respect to the acoustic patterns of individual calls (Syka et al. 1997). In this study, 4 typical guinea pig calls were used. They differ in duration, number of phrases, frequency range, and the harmonic versus noisy character.
The different call features are reflected in the neural responses. From the results, a distributed representation of whistle by a population of neurons is evident, in which neurons with different CFs responded with fundamentally different response patterns and, on the other hand, quite identical response patterns were obtained irrespective of unit CF for the other 3 calls, chirp, purr, and chutter. As demonstrated by this example, the use of more than just one call can provide a more detailed view of the coding schema, which can be masked by some effect resulting from the acoustic pattern of a particular call.
Subnuclei and unit diversity within the IC
The IC is not a homogeneous structure; 3 subnuclei have been recognized (Aitkin 1986), and also several unit types were described according to neuronal morphology and/or response properties (e.g., Oliver and Morest 1984; Ramachandran et al. 1999; Rees et al. 1997; Syka et al. 2000).
Syka et al. (2000) evaluated the responses evoked by tonal stimulation at the CF of guinea pig IC neurons and demonstrated that CNIC neurons were characterized by lower thresholds, a shorter latency, a higher rate of spontaneous and driven activity, and a sharper frequency tuning (as expressed by higher Q10 values) compared with DCIC and ECIC neurons. The authors concluded that the CNIC belongs to the fast, frequency-tuned, low threshold, and intensity-sensitive ascending pathway, whereas the other 2 IC subdivisions are involved in additional processing of information that includes feedback loops and polysensory pathways. It may be expected that the poorer frequency selectivity (lower Q10 values) found in the ECIC and DCIC compared with that in the CNIC will result in larger responses to broad-band stimuli, such as noise or vocalizations, in the ECIC and DCIC. This assumption is confirmed by the fact that tonal stimuli evoked a significantly stronger average maximal response in the CNIC than in the ECIC, but the response to noise was slightly, but not significantly, lower in the CNIC than in the DCIC and ECIC. The efficacy of complex sounds was demonstrated by Aitkin et al. (1994) in the IC of anesthetized cats. Vocal stimuli were more efficient in terms of higher firing rates than noise or CF stimuli in 27% of the units in the CNIC, 82% in the ECIC, and 72% in the DCIC.
In the present study, the average response to individual vocalization signals was not significantly different in central and peripheral IC subnuclei, which confirms a relative enhancement of responses to spectrally rich sounds in the peripheral subnuclei suggested by Syka et al. (2000), who found weaker responses to tonal stimuli in external nuclei relative to the central nucleus of the IC, but similar responses to broad-band noise in all nuclei. On the other hand, our study does not show stronger responses to vocalizations in the external nuclei relative to the central nucleus of the IC as found in cat by Aitkin et al. (1994).
Moreover, neurons within the IC are not a homogeneous population. Two groups of units in the CNIC have been distinguished according to unit morphology (Oliver and Morest 1984): disc-shaped and stellate. Units are usually divided into at least 2 groups according to their PSTH patterns, onset and sustained, but often even more categories of PSTHs have been recognized (Rees et al. 1997; Syka et al. 2000). The heterogeneity within the IC is also expressed in various rate-level functions (Syka et al. 2000), more general tone response maps (Davis et al. 1999), and also binaural interactions (Ramachandran et al. 1999).
Figure 11 shows fundamentally different firing patterns in units with similar CFs. The fact that both patterns showed a nonzero correlation to the near-CF energy (one is positive, the other is negative) suggests that different coding schemes are employed by each of the units. The response patterns of the units with positive correlation coefficients correspond with the monotonic rate-level functions for tonal stimuli, whereas the units with negative correlation coefficients had nonmonotonic rate-level functions. Variability in the rate-level function (monotonic, nonmonotonic, and saturated type) was previously reported in the IC of ketamine–xylazine-anesthetized guinea pigs by Syka et al. (2000).
The methods employed clearly showed coding of the spectrotemporal acoustic pattern in the IC, but the heterogeneity of the IC seems to be a limitation for averaging procedures. For example, averaging the different PSTHs shown in Fig. 11 could lead to information being erased rather than provided. This implies that the neurons in the IC could carry more information about the call features that seems to be the case from the presented results. As possible solutions to this problem, more sophisticated methods of analysis could be employed rather than just averaging the driven rate (Rieke et al. 1997), or neurons could be grouped into many subpopulations (Wang et al. 1995) according to their response properties. The 2nd approach is also supported by the existence of different coding schemes employed by neurons that have been reported at various levels of the auditory pathway ranging from the brain stem to the cortex. The analysis of individual neuronal types revealed separate rate and temporal representation schemes in the cochlear nucleus of cats (Blackburn and Sachs 1990); other studies of mammalian auditory brain stem circuitry have revealed parallel projections important for both sound localization and general acoustic feature extraction (Young 1998a,b). Recently, distinct subpopulations of neurons were described on the basis of the representation of spectral and temporal features of a sound also in the auditory cortex. Lu et al. (2001) reported 2 distinct populations of neurons with temporal and rate codes of time-varying signals in the auditory cortex of awake primates. In the study of Barbour and Wang (2003), the authors evaluated neurons on the basis of their sensitivity to the spectral contrast of acoustical stimuli and described low-contrast versus high-contrast preferring neurons in the auditory cortex of awake primates.
Effect of anesthesia
The fact that the presented data were recorded in anesthetized guinea pigs raises the question as to how representative are the data for awake animals. Studies performed in awake and anesthetized animals demonstrated an effect of anesthesia on the basic response properties of central auditory neurons (e.g., Gaese and Ostwald 2001; Kisley and Gerstein 1999).
The spontaneous activity of neurons reported in this study (12.3 ± 16.8 spikes/s) is similar to the average spontaneous activity of 13.3 spikes/s reported in an extensive study of the IC under identical anesthesia by Syka et al. (2000), and a similar value (17.8 spikes/s) was reported by Terterolo et al. (2002) in awake guinea pigs. Pentobarbital, compared to ketamine–xylazine anesthesia (Astl et al. 1996), reduces the spontaneous activity relative to the awake state (Terterolo et al. 2002).
Also, similar latencies and the sustained pattern as the most frequent type of response pattern to tone bursts were found in both awake (Terterolo et al. 2002) and ketamine–xylazine anesthetized animals (Syka et al. 2000). The frequency tuning in the IC evaluated on the basis of Q10 values was reported by Astl et al. (1996) to be similar, irrespective of the type of anesthesia (pentobarbital, urethane, ketamine–xylazine).
In summary, this study does not prove the existence of highly specific units in the IC, but provides evidence for the representation of the spectral and temporal features of complex sounds in a distributed way by a population of IC units. We can assume that relatively detailed information about sound features is provided by the IC neurons to higher, thalamocortical structures of the auditory pathway.
|
DISCLOSURES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: J.Syka, Institute of Experimental Medicine AS CR, Vídeská 1083, 142 20 Prague, Czech Republic (E-mail: syka{at}biomed.cas.cz).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONDISCLOSURESACKNOWLEDGMENTSREFERENCES |
|---|
Aitkin L, Tran L, and Syka J. The responses of neurons in subdivisions of the inferior colliculus of cats to tonal, noise and vocal stimuli. Exp Brain Res 98: 53–64, 1994.[Web of Science][Medline]
Astl J, Popelá J, Kvaák E, and Syka J. Comparison of response properties of neurons in the inferior colliculus of guinea pigs under different anesthetics. Audiology 35: 335–345, 1996.[Web of Science][Medline]
Barbour DL and Wang X. Temporal coherence sensitivity in auditory cortex. J Neurophysiol 88: 2684–2699, 2002.
Barbour DL and Wang X. Contrast tuning in auditory cortex. Science 299: 1073–1075, 2003.
Blackburn CC and Sachs MB. The representations of the steady-state vowel sound /e/ in the discharge patterns of cat anteroventral cochlear nucleus neurons. J Neurophysiol 63: 1191–1212, 1990.
Chiu TW and Poon PWF. Basic response determinants of single neurons to amplitude modulation in the auditory midbrain. Exp Brain Res 134: 237–245, 2000.[Web of Science][Medline]
Creutzfeldt O, Hellweg FC, and Schreiner C. Thalamocortical transformation of responses to complex auditory stimuli. Exp Brain Res 39: 87–104, 1980.[Web of Science][Medline]
Davis KA, Ramachandran R, and May BJ. Single-unit responses in the inferior colliculus of decerebrate cats. II. Sensitivity to interaural level differences. J Neurophysiol 82: 164–175, 1999.
Doupe AJ and Konishi M. Song-selective auditory circuits in the vocal control system of the zebra finch. Proc Natl Acad Sci USA 88: 11339–11343, 1991.
Esser KH, Condon CJ, Suga N, and Kanwal JS. Syntax processing by auditory cortical neurons in the FM-FM area of the mustached bat Pteronotus parnellii. Proc Natl Acad Sci USA 94: 14019–14024, 1997.
Gaese BH and Ostwald J. Temporal coding of amplitude and frequency modulation in the rat auditory cortex. Eur J Neurosci 7: 438–450, 1995.[Web of Science][Medline]
Gaese BH and Ostwald J. Anesthesia changes frequency tuning of neurons in the rat primary auditory cortex. J Neurophysiol 86: 1062–1066, 2001.
Gehr DD, Komiya H and Eggermont JJ. Neuronal responses in cat primary auditory cortex to natural and altered species-specific calls. Hear Res 150: 27–42, 2000.[Web of Science][Medline]
Goldberg JM and Brown PB. Response of binaural neurons of dog superior olivary complex to dichotic tonal stimuli: some physiological mechanisms of sound localization. J Neurophysiol 32: 613–636, 1969.
Harper LV. Behavior. In: The Biology of the Guinea Pig, edited by Wagner JE and Manning PJ. New York: Academic Press, 1976, p. 31–51.
Kao MC, Poon PW, and Sun X. Modeling of the response of midbrain auditory neurons in the rat to their vocalization sounds based on FM sensitivities. Biosystems 40: 103–109, 1997.[Web of Science][Medline]
Kisley MA and Gerstein GL. Trial-to-trial variability and state-dependent modulation of auditory-evoked responses in cortex. Neuroscience 19: 10451–10460, 1999.
Lu T, Liang L, and Wang X. Temporal and rate representation of time-varying signals in the auditory cortex of awake primates. Nat Neurosci 4: 1131–1138, 2001.[Web of Science][Medline]
Manley JA and Muller-Preuss P. Response variability of auditory cortex cells in the squirrel monkey to constant acoustic stimuli. Exp Brain Res 32: 171–180, 1978.[Web of Science][Medline]
Margoliash D. Acoustic parameters underlying the responses of song-specific neurons in the white-crowned sparrow. Neuroscience 3: 1039–1057, 1983.[Abstract]
Nagarajan SS, Cheung SW, Bedenbaugh P, Beitel RE, Schreiner CE, and Merzenich MM. Representation of spectral and temporal envelope of twitter vocalizations in common marmoset primary auditory cortex. J Neurophysiol 87: 1723–1737, 2002.
Neuert V, Pressnitzer D, Patterson RD, and Winter IM. The responses of single units in the inferior colliculus of the guinea pig to damped and ramped sinusoids. Hear Res 159: 36–52, 2001.[Web of Science][Medline]
Newman JD. Perception of sounds used in species-specific communication: the auditory cortex and beyond. J Med Primatol 7: 98–105, 1978.[Web of Science][Medline]
Newman JD and Symmes D. Arousal effects on unit responsiveness to vocalizations in squirrel monkey auditory cortex. Brain Res 78: 125–138, 1974.[Web of Science][Medline]
Newman JD and Wollberg Z. Multiple coding of species-specific vocalizations in the auditory cortex of squirrel monkeys. Brain Res 54: 287–304, 1973.[Web of Science][Medline]
Ohlemiller KK, Kanwal JS, and Suga N. Facilitative responses to species-specific calls in cortical FM-FM neurons of the mustached bat. Neuroreport 7: 1749–1755, 1996.[Web of Science][Medline]
Oliver DL and Morest DK. The central nucleus of the inferior colliculus in the cat. J Comp Neurol 222: 237–264, 1984.[Web of Science][Medline]
Pelleg-Toiba R and Wollberg Z. Discrimination of communication calls in the squirrel monkey: "call detectors" or "cell ensembles"? J Basic Clin Physiol Pharmacol 2: 257–272, 1991.[Medline]
Poon PWF and Chiu TW. Single cell responses to AM tones of different envelops at the auditory midbrain. In: Acoustical Signal Processing in the Central Auditory System, edited by Syka J. New York: Plenum Press, 1997, p. 253–261.
Ramachandran R, Davis KA, and May BJ. Single unit responses in the inferior colliculus of decerebrate cats. I. Classification based on frequency response maps. J Neurophysiol 82: 152–163, 1999.
Rauschecker JP, Tian B, and Hauser M. Processing of complex sounds in the macaque nonprimary auditory cortex. Science 268: 111–114, 1995.
Rees A and Moller AR. Responses of neurons in the inferior colliculus of rat to AM and FM tones. Hear Res 10: 301–330, 1983.[Web of Science][Medline]
Rees A and Palmer AR. Neuronal responses to amplitude-modulated and pure-tone stimuli in the guinea pig inferior colliculus, and their modification by broadband noise. J Acoust Soc Am 85: 1978–1994, 1989.[Web of Science][Medline]
Rees A, Sarbaz A, Malmierca MS, and Le Beau FE. Regularity of firing of neurons in the inferior colliculus. J Neurophysiol 77: 2945–2965, 1997.
Rieke F, Warland D, van Steveninck RD, and Bialek W. Spikes: exploring the neural code. Cambridge, MA: MIT Press, 1997.
Shannon RV, Zeng FG, Kamath V, Wygonski J, and Ekelid M. Speech recognition with primarily temporal cues. Science 270: 303–304, 1995.
Syka J, Popelá J, Kvaák E, and Astl J. Response properties of neurons in the central nucleus and external and dorsal cortices of the inferior colliculus in guinea pig. Exp Brain Res 133: 254–266, 2000.[Web of Science][Medline]
Syka J, Popelá J, Kvaák E, uta D, and Jilek M. Processing of species-specific vocalizations in the inferior colliculus and medial geniculate body of the guinea pig. In: Acoustical Signal Processing in the Central Auditory System, edited by Syka J. New York: Plenum Press, 1997.
Tanaka H and Taniguchi I. Responses of medial geniculate neurons to species-specific vocalized sounds in the guinea pig. Jpn J Physiol 41: 817–829, 1991.[Web of Science][Medline]
Torterolo P, Falconi A, Morales-Cobas G, and Velluti RA. Inferior colliculus unitary activity in wakefulness, sleep and under barbiturates. Brain Res 935: 9–15, 2002.[Web of Science][Medline]
Wang X and Kadia SC. Differential representation of species-specific primate vocalizations in the auditory cortices of marmoset and cat. J Neurophysiol 86: 2616–2620, 2001.
Wang X, Merzenich MM, Beitel R, and Schreiner CE. Representation of a species-specific vocalization in the primary auditory cortex of the common marmoset: temporal and spectral characteristics. J Neurophysiol 74: 2685–2706, 1995.
Wenstrup JJ and Grose CD. Inputs to combination-sensitive neurons in the medial geniculate body of the mustached bat: the missing fundamental. Neuroscience 15: 4693–4711, 1995.[Abstract]
Wenstrup JJ, Mittmann DH, and Grose CD. Inputs to combination-sensitive neurons of the inferior colliculus. J Comp Neurol 409: 509–528, 1999.[Web of Science][Medline]
Winter P and Funkenstein HH. The effect of species-specific vocalization on the discharge of auditory cortical cells in the awake squirrel monkey (Saimiri sciureus). Exp Brain Res 18: 489–504, 1973.[Web of Science][Medline]
Wollberg Z and Newman JD. Auditory cortex of squirrel monkey: response patterns of single cells to species-specific vocalizations. Science 175: 212–214, 1972.
Young ED. Parallel processing in the nervous system: evidence from sensory maps. Proc Natl Acad Sci USA 94: 933–934, 1998a.
Young ED. Cochlear nucleus. In: The Synaptic Organization of the Brain, edited by Shepherd GM. New York: Oxford Univ. Press, 1998b.
Young ED and Sachs MB. Representation of steady state vowels in the temporal aspects of the discharge patterns of population of auditory-nerve fibers. J Acoust Soc Am 66: 1381–1403, 1979.[Web of Science][Medline]
This article has been cited by other articles:
|
|
 |
|
  K. A. Razak, M. D. Richardson, and Z. M. Fuzessery Experience is required for the maintenance and refinement of FM sweep selectivity in the developing auditory cortex PNAS, March 18, 2008; 105(11): 4465 - 4470. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. A. Razak and Z. M. Fuzessery Development of Inhibitory Mechanisms Underlying Selectivity for the Rate and Direction of Frequency-Modulated Sweeps in the Auditory Cortex J. Neurosci., February 14, 2007; 27(7): 1769 - 1781. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. Gourevitch and J. J. Eggermont Spatial Representation of Neural Responses to Natural and Altered Conspecific Vocalizations in Cat Auditory Cortex J Neurophysiol, January 1, 2007; 97(1): 144 - 158. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. DiMattina and X. Wang Virtual Vocalization Stimuli for Investigating Neural Representations of Species-Specific Vocalizations J Neurophysiol, February 1, 2006; 95(2): 1244 - 1262. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Xie, J. Meitzen, and G. D. Pollak Differing Roles of Inhibition in Hierarchical Processing of Species-Specific Calls in Auditory Brainstem Nuclei J Neurophysiol, December 1, 2005; 94(6): 4019 - 4037. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Cedolin and B. Delgutte Pitch of Complex Tones: Rate-Place and Interspike Interval Representations in the Auditory Nerve J Neurophysiol, July 1, 2005; 94(1): 347 - 362. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
